vibehacker
News
GitHub / Benzi ·

Benzi: compiler-backed coding agent hits 78.2% on SWE-bench Verified for $37

Benzi parses your repo with tree-sitter into a resolved map of symbols, call edges, data flow, and inheritance, then has the model query it through ~36 structured tools instead of reading raw files; it ships as a VS Code extension, a pip install benzi package, and an MCP server. Its authors report 391/500 on SWE-bench Verified with DeepSeek v4-flash at pass@1 for $37.33 total, a self-run result that includes a review pass, not just the index.

More news

View all

Show HN: Loa, a local-first coding agent that plans work as a mutable DAG

Loa runs long coding tasks against your own OpenAI compatible endpoint (Ollama works), building acceptance criteria and a replannable DAG, then rebuilding context from state each step instead of compressing chat history; the author suggests 27B–32B local models. It makes many model calls by design, so skip metered cloud APIs, and use the optional Docker sandbox since host mode runs shell commands with your privileges…

Show HN / GitHub

Curie: open-source platform to ship Claude Code-style agents to Slack and Kubernetes

Curie (Apache 2.0, v0.12.0) takes a Claude Code format plugin bundle (skills, tools, MCP) and runs the same snapshot as a single container, a local Docker Compose stack, or a Kubernetes release, with traces, evals, budgets, and deploys on git push . Slack is the first channel, with email and Teams next; models can be Anthropic, OpenRouter, or local Ollama, and curie try gets a first reply without credentials…

GitHub / Curie

AWS devs open-source Pizza Bot: a local-first inbox for background AI agents

Pizza Bot (Apache 2.0, a community project rather than an AWS service) runs scheduled or webhook triggered agent work in the background, delegates to specialist agents, and pauses in an Action queue for your approval before consequential steps. It supports multiple model providers, MCP servers, and Agent Skills, and keeps threads, checkpoints, and logs in a local folder with credentials in the OS secret store…

InfoQ

Spotted something we missed? Start a thread.